## Group Research

### Mohan Chen

#### How to use orbital-free density functional theory to study liquid metals?

Orbital-free density functional theory (OFDFT) is a powerful linear-scaling method for electronic structure calculations. Instead of calculating individual Kohn-Sham (KS) orbitals, as in KS-DFT, OFDFT uses the electron density as the sole working variable, and, thus, is capable of simulating large systems that contain millions of atoms. With the recent progress in OFDFT methods, OFDFT is no longer restricted to light metals, but can also be applied to semiconductors, transition metals, or molecules. In my study, I will carry out *ab-initio* molecular dynamics (MD) to investigate various properties of liquid metals. These properties are used as important plasma-facing materials in fusion reactors. The MD simulations will also be combined with newly developed techniques in OFDFT, namely small-box techniques. While traditional fast Fourier transform (FFT) implementations scale poorly for more than a few hundred processors, the newly developed small-box FFT enables us to do highly parallelized calculations using several tens of thousands of cores to do FFT. Small-box FFT will also allow us to carry out MD simulations on several tens of thousands of atoms. (Back)

### Jin Cheng

#### How to combine CW methods and DFT methods for higher efficiency and accuracy?

A quantitative understanding of a chemical process requires a balanced consideration of efficiency and accuracy. Density functional theory (DFT) scales well with system size, and can readily describe extended systems by the use of periodic boundary conditions, to obtain, e.g., bulk properties. Unfortunately, DFT is limited to ground-state calculations and the use of approximate exchange-correlation functionals. More sophisticated ab-initio correlated wavefunction (CW) methods feature a better description of electronic correlation and excited states. However, the prohibitively large computational cost severely limits their applicability. Furthermore, combining the exact treatment of correlation with the use of periodic boundary conditions is challenging. One possible strategy to overcome these restrictions is to partition a larger system into parts of manageable size that can be treated on different levels of theory: one describes the region of interest (cluster) via a CW method, while the environment is treated with DFT. Such an embedding scheme should reproduce the electronic structure of the cluster while reducing the computational costs to an acceptable level. Our group has developed a potential-functional-based embedding theory [ *J. Chem. Phys.*, **135**, 194104 (2011)] for this purpose, that (1) avoids ambiguities by introducing a global embedding potential to model the interaction between subsystems, (2) allows for flexible combination of different methods (DFT, CW), and (3) can be done in a self-consistent fashion. Embedding applications include the interaction of gas molecules with metal surfaces (CO on copper, O_{2} on Aluminum), that require both a high-level treatment at the adsorption site as well as a description of the extended surface. In the future, the CW/DFT embedding scheme will be used to study the mechanism of a chemical process, e.g., the oxygen reduction reaction on the cathode material of solid oxide fuel cells. (Back)

### Martina Lessio

#### How can we convert CO_{2} into liquid fuels and chemicals?

Identifying renewable energy sources and reducing atmospheric CO_{2} require immediate attention. Photocatalytic reduction of CO_{2} to useful products may handle both of these issues. The aim of my research is to study this catalytic process using first principles quantum mechanics methods. In particular I will consider the semiconductor gallium phosphide (GaP) as a photocatalyst. I will investigate the interactions between its surface and the species participating in the CO_{2} reduction. In addition to the solid catalyst, recent studies suggest that a co-catalyst formed from a pyridinium ion might play a crucial role in achieving efficient CO_{2} reduction over the semiconductor surface. As part of my research, I will work on elucidating the reaction mechanism involving the pyridinium ion, CO_{2}, water, and the GaP surface that leads to CO_{2 }reduction. (Back)

### Ting Tan

#### How do oxygenated biofuels perform differently from conventional hydrocarbon based fuels?

The typical feature of biofuels which distinguishes them from conventional hydrocarbon fuels is the oxygen atoms included as an additional element in the molecular constitution. The presence of the oxygen atoms in biofuels changes the properties of these fuel molecules, and then makes their combustion behavior different from familiar hydrocarbon fuels. Their chemical decomposition and oxidation pathways are well coupled to the structure of the corresponding fuel molecules. Predicting the combustion behavior of these fuels requires the development of detailed combustion mechanisms, which must include quite a large number of reactions, rate coefficients, as well as related thermochemical and transport parameters. I am using the ab initio multireference configuration interaction method to find out the effect of these oxygen atoms on the fuel molecules, and to improve the accuracy of the parameters which are necessary to define the combustion mechanism. (Back)